At the 2025 Journal of Accounting and Public Policy Conference, a panel of three senior internal audit leaders describe how internal audit has shifted from a narrow, compliance-centric function to a broad, technology-enabled assurance and advisory partner. This transformation has been propelled by significant public policy developments in 2002, 2016, and 2023, and most recently by generative artificial intelligence (GenAI). Panelists described parallel journeys: professionalizing internal audit; embedding IT audit, data analytics, automation, and AI governance “guardrails”; and retooling teams to include data scientists and “digital auditors”. They also discussed interactions with external auditors and regulators, which impact how to test, rely on, and report AI-enabled work. Looking ahead, panelists expect internal audit to (1) use GenAI to radically increase coverage, speed, and quality; (2) audit GenAI across the enterprise; (3) advise on GenAI policy and governance; and (4) help organizations navigate emerging external GenAI regulations. The panel underscored the importance of cultivating professional skepticism amid structural threats from automation and remote work. Emerging from the panel discussion are timely and relevant public policy issues on the use of GenAI in internal and external audits, and interesting research opportunities.
This study examines cross-industry labor market-based information transfer in the setting of quarterly earnings conference calls held by S&P 500 firms from 2002 to 2019. We find a significant association between the abnormal stock returns of the announcing firms' labor market peers and human capital-related information from the conference calls after controlling product market-based information transfer and other confounding factors. Our study extends the literature from product market-based to labor market-based and from intra-industry to cross-industry information transfers and highlights human capital disclosures in conference calls as an important channel through which information is transmitted across peer firms.
We examine the informativeness of voluntary disclosure of web traffic metrics (i.e., the active users and customers) by internet firms. We show that web traffic disclosures, adopted by approximately 30 percent of the firms in our sample, are relevant for predicting future sales growth, useful to analysts when making future sales forecasts, and priced by investors at earnings announcements. In particular, our findings are obtained after controlling for similar metrics provided by third-party vendors on a more frequent basis. This suggests that firms' web traffic disclosures (which serve as a key non-financial performance indicator for internet firms) contain incremental information beyond third-party estimates. In additional analyses, we explore the characteristics of firms that voluntarily disclose web traffic metrics and examine whether they exhibit greater reporting opportunism than their non-disclosing counterparts.
We investigate stakeholder perceptions of events leading up to the introduction of the first standard for tax-related sustainability reporting, Global Reporting Initiative (GRI) 207. Issued by the Global Sustainability Standards Board, GRI is the dominant framework worldwide for voluntary sustainability reporting. First, we examine the perceptions of a broad group of stakeholders by analyzing comment letter responses to the GRI 207 exposure draft, finding that a majority of these responses support the standard and highlight increasing demand for tax transparency. Next, we focus on preparers of tax transparency reports by conducting qualitative interviews. Interviewees commented on the increasing pressure to provide tax information and concerns about implementation and proprietary costs, as well as skepticism about the benefits. Finally, we use an event study to examine investor perceptions in which we document significant negative average cumulative abnormal returns surrounding events surrounding the adoption of GRI 207 for a sample of European GRI-reporting firms. We interpret this result as evidence that investors perceive the net costs of tax transparency under the new standard to be greater than the benefits. Our findings reveal varying viewpoints on the benefits and costs of tax transparency, while underscoring the broader public-interest implications for policy and professional practice.
This study examines whether the aggregation of domestic geographic disclosures influences corporate tax avoidance in China. We find a positive and statistically significant association between domestic geographic segment aggregation and tax avoidance. Our results are robust to alternative measures of domestic geographic segment aggregation and various endogeneity tests. Cross-sectional analyses show that the positive association is stronger for firms with weak tax enforcement. Finally, mechanism analyses indicate that higher levels of domestic geographic disclosure aggregation are associated with greater tax avoidance, both directly and indirectly through the domestic-to-domestic income shifting channel. This study extends our understanding of overseas geographic segment disclosure aggregation and tax avoidance by offering new evidence on the role of domestic geographic disclosure aggregation. Our study should also inform policymakers involved in debates about financial reporting standards and the use of domestic geographic information disclosures in financial reports.
This study examines whether a firm’s excessive or insufficient changes in eXtensible Business Reporting Language (XBRL) tags affect the over time comparability of financial information and, in turn, influence analyst forecasts. We measure excessive or insufficient changes by abnormal changes in XBRL tags, defined as the absolute values of residuals from a model that regresses a firm’s year-over-year difference in XBRL tags on changes in related financial reporting factors. Using a sample from 2013 to 2020, we find that firms with more abnormal changes in XBRL tags, particularly those with excessive changes, exhibit lower analyst coverage, higher forecast errors, and greater forecast dispersion. We also find that abnormal changes in standard XBRL tags have a stronger effect on analyst forecasts than those in custom tags. Our study provides important insights for financial report users and regulators, particularly given that the U.S. Securities and Exchange Commission (SEC) has recognized the issue of excessive changes in XBRL tags.
This paper studies the role of conservative accounting in motivating agents who compare and care about their performance information. I develop a moral-hazard model with a principal and two agents, in which the principal chooses the level of conservatism and each agent may gain or lose utility when his performance is higher or lower, respectively, than that of his peer. I show that the principal sets an interior level of conservatism when the agents are sufficiently sensitive to performance comparison and the task is challenging enough. This result provides a novel rationale for accounting conservatism.
Amid recent attention over auditors’ ability to properly assess client firms’ going concerns, this study examines the potential for auditor and Machine Learning (ML) collaboration in assessing going concerns. We find that while ML generates fewer overall errors than auditors’ Going Concern Opinions (GCOs), particularly in reducing false positives, replacing auditors with ML is not a superior solution, as each tends to make errors on different cases. Instead, integrating the strengths of both approaches may help offset their respective limitations. Using an archival approach, we derive a “team” assessment that integrates auditors and ML. Our results show that the team assessments can synergize the relative advantages of auditor and ML, and overcome their relative disadvantages, serving as a potential way of generating higher quality going concern assessments. Our research has implications for accounting practitioners and regulators regarding auditor-machine collaboration in the “AI era.”
The increasing adoption of Artificial Intelligence (AI) technologies in accounting practices is reshaping traditional processes, offering unprecedented opportunities while also posing significant challenges for organizations. However, previous research has primarily focused on large accounting firms, leaving AI adoption patterns in non-accounting firms, where 76% of accountants are employed, largely underexplored. This study addresses this gap by conducting the first comprehensive comparative analysis of AI adoption, uncovering novel insights into levels, drivers, and barriers of AI integration and how they differ between accounting and non-accounting firms. Based on 35 semi-structured interviews, the study reveals an interplay between organizational-level and individual-level factors that shape adoption of different types of AI: robotic process automation (RPA), analytical AI (i.e., machine learning), and generative AI (i.e., large language models (LLMs)). It applies the Technology-Organization-Environment (TOE) framework as analytical lens and extends it by incorporating an individual dimension drawn from the UTAUT framework. The analysis reveals distinct patterns. Accounting firms predominantly employ RPA and ML for assurance-driven tasks such as transaction testing, anomaly detection, and regulatory compliance, while non-accounting firms adopt AI largely for predictive financial forecasting and process optimization. Key barriers, including rigid business models in accounting firms and organizational silos in non-accounting firms shape adoption trajectories. This study contributes to the field of accounting and auditing research by offering a nuanced understanding of how organizational context and professional logics uniquely influence AI adoption. It provides actionable insights for academics, practitioners, and policymakers, emphasizing the need for cross-functional collaboration, regulatory frameworks, and AI literacy initiatives to foster effective adoption across diverse sectors.
This paper examines stock market reactions to the Silicon Valley Bank (SVB) and Signature Bank (SB) failures in March 2023. Using an event study of U.S. bank holding companies, we document significant negative abnormal returns surrounding the failures, with losses emerging prior to the SVB closure and intensifying on the event dates. We further analyze cross-sectional heterogeneity in market reactions based on banks’ common exposures to SVB and SB. Banks with similar balance sheet characteristics—particularly large holdings of held-to-maturity and available-for-sale securities, sizable unrealized losses, concentrated lending portfolios, and high uninsured deposits—experienced significantly more adverse stock price responses. These findings are consistent with an indirect contagion channel in which investors react to common unfavorable signals rather than direct interbank linkages. Overall, the results inform ongoing policy debates regarding accounting measurement, disclosure, and banking sector stability during periods of systemic stress.
We investigate whether and how investors respond to firm-level macroeconomic risk awareness (FMRA) through cost of equity capital (COE) against the backdrop of increased macroeconomic risk. We develop a measure of FMRA by applying Bidirectional Encoder Representations from Transformers (BERT) to analyze the texts of mandatory Management Discussion and Analysis (MD&A) sections of annual reports. For a sample of 16,949 firm-year observations from 3,303 Chinese listed firms during the period 2010-2020, we document that FMRA weakens the positive relation between firms' macroeconomic risk exposure and COE. Besides, we find that reduced operating risk, narrowed information asymmetry and mitigated agency risk play important roles in the contingent effect of FMRA. Further, leveraging the awareness-motivation-capability (AMC) framework, our findings suggest that the contingent effect of FMRA is only significant for firms facing high stakeholder pressure and those with high levels of unabsorbed organizational slack. Our findings also suggest that FMRA contributes to greater post-shock stability, one key aspect of organizational resilience. Our study highlights the importance of FMRA in reducing firm risk and the capital market appears to incorporate forward-looking indicators of firm risk in investment decision-making.
This study examines the relation between blockchain adoption and investment efficiency. Using a difference-in-differences approach with a matched sample of firms that disclosed blockchain adoption in 8-K filings between 2014 and 2023, we find that adopters exhibit higher investment efficiency after adoption compared to non-adopters. We also find that blockchain adoption is positively associated with financial reporting quality, suggesting that enhanced financial reporting quality might be a mechanism through which blockchain adoption improves investment efficiency. Further analyses show that the effect of blockchain adoption on investment efficiency varies with firms’ information environments and across types of blockchain applications. The effect is most pronounced among firms with weak information environments and those adopting non-cryptocurrency applications, where blockchain’s potential to improve investment efficiency is expected to be greater. Overall, our findings provide insights into how blockchain technology can influence financial reporting practices and corporate investment decisions.
This study examines the association of the Tax Cuts and Jobs Act (TCJA) and innovation by analyzing changes in R&D expenditures and patent outputs. I focus on two key events: the enactment of the law in December 2017 and the enforcement of the capitalization and amortization (C&A) requirement in January 2022. The results show a 6.63% annual decrease in R&D expenditures before enforcement and a 3.31% decrease post-enforcement, with the manufacturing and high-tech sectors being most affected. The number of patents granted declined significantly, challenging the notion that firms merely reclassified R&D expenses to capitalize on tax benefits while maintaining innovation. Controls for COVID-19 effects and other confounding factors confirm the robustness of these findings. Finally, I find negative relations between innovation cycles and R&D expenses after the enactment date. The findings highlight how firms responded to various provisions of the TCJA, including the C&A requirement, and their implications for innovation and economic growth, providing critical insights for policymakers and researchers at the intersection of tax policy and accounting practices.
We examine how low financial statement comparability with industry peers affects textual disclosures—specifically length, complexity, forward-looking content, and hard information (presented by the number of tables)—in the Management Discussion and Analysis (MD&A) and in the notes to the financial statements of 10-K filings. We find that lower comparability with industry peers is associated with higher information quantity and complexity in the MD&A and the notes to the financial statements. Moreover, we discover that the quantity of forward-looking information in the MD&A, as well as numerical information in the notes to the financial statements, is higher when financial statement comparability is lower. Capital market tests show that adjusted disclosures are associated with reduced information asymmetry between managers and investors, greater analyst following, and lower analyst dispersion. Furthermore, lower financial statement comparability is associated with a higher frequency of voluntary disclosures and a higher quantity of disclosures in 8-K filings. Overall, our findings support the argument that, when financial statement comparability is low and, consequently, information asymmetry between firms and market participants is heightened, managers adjust their textual disclosures to reduce information asymmetry.
One in seven Chinese politicians has prior CEO experience, governing jurisdictions that account for about thirty percent of listed firms. This study investigates the impact of municipal politicians with prior CEO experience (PCEs) on corporate investment decisions. We find that the presence of PCEs is associated with lower corporate investment efficiency. This inefficiency is particularly pronounced when PCEs have longer prior CEO tenures, come from larger firms or industries with a stronger local presence, originate from private firms, or face stronger promotion incentives and greater political power. Additional analyses indicate that PCE-led cities exhibit higher GDP growth, and PCEs are more likely to be promoted than their counterparts without CEO experience. While firms in PCE-led cities benefit from increased financial support, such as government subsidies and loans, their long-term financial performance deteriorates. Collectively, this study provides systematic evidence that local officials strategically leverage their professional expertise to stimulate economic growth, but at the cost of corporate investment efficiency.
PCAOB inspection reports provide public insights into audit quality and often shape perceptions of auditor credibility. Drawing on source credibility theory, we examine how expertise (deficiency rates) and trustworthiness (breaches of trust) influence judgments by non-professional investors. Across two experiments, we find that higher deficiency rates and breaches, whether firm-level (subverting inspections) or engagement-specific (improperly revising materiality), reduce perceptions of auditor due care, financial statement accuracy, and investment likelihood, with the firm-level breach amplifying the effect of low expertise. Supplementary analyses reveal that individuals high in Machiavellian traits interpret breaches as investment opportunities, diverging from predictions based on source credibility. A third experiment demonstrates that firm responses emphasizing accountability and corrective action mitigate but do not eliminate the negative consequences of breaches, especially when expertise is perceived as low. These findings extend research on audit quality and source credibility and provide policy-relevant insights for regulators, audit committees, and audit firms about maintaining trust in capital markets.
This study examines the impact of government regulatory oversight of publicly listed firms on the stock market’s valuation of corporate cash holdings. We utilize the setting of the China Securities Regulatory Commission’s (CSRC) Random Inspection Scheme (RIS), which monitors firms’ compliance with corporate disclosure and governance requirements of the CSRC, to assess whether and how regulatory oversight influences the valuation of cash holdings. We find that the market value of cash holdings increases when a firm is inspected under the RIS, with stronger effects for firms facing greater agency conflicts and higher information asymmetry between corporate insiders and outsiders. In addition, the effect is more pronounced when other external monitoring mechanisms are weaker. Overall, our results suggest that regulatory oversight could mitigate the value destruction associated with cash holdings by alleviating agency conflicts and information asymmetry between corporate insiders and outsiders.
This study examines how firms adjust their disclosure strategies for quarterly management earnings forecasts (hereafter, ‘earnings guidance’) across the several stages of the corporate life cycle. We find that firms are more likely to initiate earnings guidance during the growth stage and maintain it through the mature stage, whereas firms in the introduction, shakeout, and decline stages are more likely to withhold or discontinue offering such guidance. Such a reversal in guidance behavior across life-cycle stages reflects changes in the relative benefits and costs of disclosure. Growth firms demonstrate greater accuracy and credibility in their earnings guidance, effectively guiding analysts toward beatable earnings targets. In contrast, firms in the introduction and decline stages exhibit characteristics associated with lower benefits and higher costs for issuing guidance, including reduced accuracy in meeting their own targets and limited success in aligning with analysts’ earnings expectations. Further analysis shows that the market reacts more strongly to forecasts from growth firms, which are perceived as more credible and result in narrower bid-ask spreads. Overall, our findings highlight the life-cycle effect of firms’ voluntary disclosure strategies.
